Understanding Global Precipitation Patterns in the Era of Climate Change

Gianluca Scuri 886725

Climate change is one of the most pressing challenges of our time, profoundly impacting various aspects of our planet's ecosystems and weather patterns. Among its many consequences, the alteration of global precipitation patterns is a crucial area of concern. Changes in precipitation have far-reaching implications for agriculture, water resources, biodiversity, and human livelihoods. There is much more agreement by the models that a warming climate will increase the severity of extreme rainfall and snowfall almost everywhere. A warmer world will, they project, also increase soil evaporation and reduce snowpack, exacerbating droughts even in the absence of reduced precipitation.

In this project long-term historical precipitation data and climate models are gathered and analyzed to identify changes in precipitation patterns over time and across different regions. In particular the objective is to uncover significant changes in the amount of the precipitations in a 100 years range and changes in the frequency and intensity of extreme precipitation events of the last years.

1. Precipitation trend 1951-2050

Import datasets

GPCC monthly data (1951-2020)

The first dataset used is provided by the Global Precipitation Climatology Centre (GPCC), operated by DWD under the auspices of the World Meteorological Organization (WMO). It contains monthly land-surface precipitation data from rain-gauges built on GTS-based and historical data. In particular, contains 50-year of precipitation data from 1951 to 2020 gridded at 1° lat/lon resolutions. This dataset is designed for climate variability and trend studies.

From here can be noticed that latitude and longitude refer to cells centroid.

CMIP6 (2015-2050)

In order to predict the future trend a global climate models is used. The climate models can simulate future precipitation patterns under different climate scenarios incorporating a range of factors such as greenhouse gas emissions, atmospheric circulation, and oceanic processes. By comparing different scenarios, we can assess the potential range of changes and uncertainties associated with future precipitation.

The scenarios are defined by two factors: the Shared Socio-economic Pathways (SSPs) and Representative Concentration Pathway (RCP). The SSPs are scenarios of projected socioeconomic global changes up to 2100 and span a range from very ambitious mitigation to ongoing growth in emissions. RCPs define the stabilized radiative forcings (the difference between incoming solar radiation from the sun, and outgoing energy radiated back into space by Earth) at the end of the century.

The model used in this project comes from the Coupled Model Intercomparison Project Phase 6 (CMIP6) and is called Earth3. For this analysis two scenarios are considered: the worst case scenario with a Fossil-fueled development and high RCP (SSP5-8.5) and a best case scenario with a sustainability approach and low RCP (SSP1-1.9).

From here can be noticed that latitude and longitude refer to cells centroid but the longitude range is [0, 359]. Let's convert longitude coordinates to [-180, 179] range.

Fix units

As these are cumulative measurements, it is better to use average values of precipitation per day in order not to introduce distortions due to the number of days in the months.

GPCC precipitation unit: $[\frac{mm}{month}]$

CMIP6 precipitation unit: $[\frac{kg}{m^2 \cdot sec}]$

Let's use $[\frac{kg}{m^2 \cdot day}]$ as common unit assuming $\rho = 1000 \frac{kg}{m^3}$

Datasets exploration

GPCC

The grid cells are empty over the oceans

It is evident an oscillation of the values following the months, let's investigate this further dividing the two hemispheres contributes.

This plot shows that the precipitation cycles for the two hemispheres are in antiphase (please note that these plots are obtained with an arithmetic mean and not with a mean weigthed on the cells areas so the amplitude is incorrect).

Is evident that the peak of precipitation shifts between the two hemispheres. Let's now check that the cells used are always the same for every month.

Let's import a shapefile with the boarders of the world countries to locate the grid cells.

Different locations show very different precipitation patterns:

CMIP6

The difference with the observation dataset (GPCC) is that here we have a land + oceans coverage

The two scenarios shows some differences on the location of the precipitations but a really small difference in the total precipitation budget

Regridding

The blue grid corresponding to the Earth3 dataset and has a lower resolution. Let's regrid the gpcc dataset (orange grid) according to the Earth3 grid

Let's use the nearest interpolation because preserve the global mean precipitation value

Keep only the common cells

Precipitation normals (1961-1990)

The climate normals serve as a benchmark against which recent or current observations can be compared, including providing a basis for many anomaly based climate datasets. The period from 1961 to 1990 has been retained as a standard reference period for long-term climate change assessments (WMO Guidelines on the Calculation of Climate Normals - 2017 edition)

Monthly anomalies

Arithmetic vs weighted mean

There are great differences in values between the arithmetic and weighted mean regarding the precipitation. Although the differences are smaller in terms of anomalies let's use the weighted one when an average along the latitude axis is performed.

Compare the common period

Let's compare the period from 2015 and 2020 which is the only in common between the two datasets. This check is necessary because the precipitation normals are going to be calculated only on the GPCC dataset and used also for the models.

Precipitations

Anomalies

The pvalues are smaller than 0.05 so we reject the null hypothsis, so the differences are statistically significant. There are many possible reason that can explain these differences. Regarding the GPCC dataset the major error sources affecting gridded precipitation estimates based on rain gauge measurements are:

  1. The systematic gauge-measuring error resulting from evaporation out of the rain gauge and aerodynamic effects, when droplets or snow flakes are drifted by the wind across the gauge funnel
  2. The sampling error depending on the network density

The systematic gauge-measuring error is generally an undercatch of the true precipitation (Sevruk 1982, 1985). Parameters affecting the efficiency of gauge measurement are features of the instrument used (size, shape, exposition, etc.) and the meteorological conditions (wind, air temperature, humidity, radiation) during the precipitation event. The precipitation phase (liquid, solid, mixed), as well as the intensity (i.e., drizzle, shower) of a precipitation event, plays an important role, too.

GPCC dataset in not bias corrected for systematic gauge measuring errors.

Spatial aggregation

As can be seen from this graph, there is a noticeable shift in mean between the values of the GPCC dataset and the CMIP6 dataset. This is mainly caused by the previously mentioned reasons.

However the interesting thing from this graph is the slight upward trend in anomalies, and subsequently in precipitation, for the entire period from 1950 to 2050 for both the datasets. In the next section, we investigate whether this increase in precipitation amount is due to an increase in extraordinary precipitation in the period 1982-2020 using daily data.

2. Extreme precipitation events 1982-2020

Extreme precipitation is related to climate change in that, all else being equal, a warmer atmosphere can “hold” more water vapor, and therefore deliver more rainfall when conditions for heavy precipitation events occur. Let's see if this phenomena is noticeable from the observational data.

Import dataset

GPCC daily data 1982-2020

This is the same dataset used in the previous section but with a daily resolution that allows to evaluate the extreme precipitation events. It contains 39 years of data from 1982 to 2020 with the same 1x1 grid.

Dataset exploration

The mean precipitation in the Southern hemisphere are greater but are located on a smaller area

Reference period

Let's compute the 95th percentile on the 30-years period 1991-2020

Indices

R95p

This climate index is a measure of heavy precipitation (N days above the 95th percentile), with high values corresponding to a high chance of flooding. An increase of this index with time means that the chance of flood conditions will increase.

Precipitation total when RR > 95ptile relative to the period 1991-2020.

The number of days with precipitations above the 95th percentile (refered to the period 1991-2020 marked in red) increased in the last 39 years by 1.5 days, so it means that heavy rainy days happened more frequently (the mean is around 18.25 which is the 5% of the 365 days). This result is in according with the previous section and is also in according with the prediction of the future precipitations.

R95pTOT

Contribution to total precipitation from very wet days (above 95th percentile).

This graph shows that the percentage is pretty stable so the increment in the global precipitation is both in the extreme events and in the normal events.